tweet_deleter.py

########### GUIDANCE
# tweet deleter that deletes all of the tweets in an archive(twitter often leaves a few of your most recent ones out of the archive- use an api based deleter to do these) from the first ones you tweeted to the latest. 
# no fancy options(like checking for highly favd tweets/etc)
# common errors: twitter won't let you unretweet tweets from accounts you don't have access to- expect a few "Sorry, you are not authorized to see this status" errors at the bottom/as you go through
# tweets that aren't deleted are shown twice: once as an error message as you progress through, and once at the bottom in a summary of all errors.
# takes about 1h/10k tweets
# hacky and ugly don't judge me
############ USAGE GUIDE
# download your twitter archive and unzip it. 
# set the archive_folder variable in the "edit this" section to the path you unzipped it to
# get twitter credentials from https://apps.twitter.com
# run and check the first few tweets are going- you should be able to see your total number of tweets dropping on your profile page in real time. 
# when it's finished, check there's no weird errors leaving tweets you want deleted behind(there shouldn't be, on a 50k test run)
# If it gets canceled, make a note of how many tweets were deleted, subtract 1, and set that to the offset, then run it again- it'll start from there
############ EDIT THIS
#leave my first offset tweets alone(useful if restarting after a sleep/network drop/etc)
offset = 0
#absolute path to the folder where you unzipped the archive twitter sent you(terminated in a /)
archive_folder= ""
#twitter api keys
consumer_key = ""
consumer_secret = ""
access_token = ""
access_secret = ""
#everything below here is hacky bullshit
import glob
import json
import tweepy
def main():
    auth()
    tweetsToDelete = getTweetIds(getFileNames(archive_folder))[offset:]
    deleteTweets(tweetsToDelete)
    announceErrors()
def progressInit(toCount):
    global noOfTweets
    global noOfTweetsDeleted
    global errors
    noOfTweets = len(toCount)
    noOfTweetsDeleted = offset
    errors = []
def announceErrors():
    for printError in errors:
        print printError['id']
        print printError['error']
def auth():
    global api
    tweepyAuth = tweepy.OAuthHandler(consumer_key, consumer_secret)
    tweepyAuth.set_access_token(access_token, access_secret)
    api = tweepy.API(tweepyAuth)
def deleteTweet(tweetID):
    api.destroy_status(tweetID)
    print (progress(tweetID))
def deleteFailed(errorIn, tweetID):
    print "failed to delete %s"%(tweetID)
    print errorIn
    global noOfTweetsDeleted
    errors.append({'id':tweetID, 'error':errorIn})
    noOfTweetsDeleted = noOfTweetsDeleted + 1
def progress(idprint):
    global noOfTweetsDeleted
    noOfTweetsDeleted = noOfTweetsDeleted + 1
    return "%s/%s %s"%(noOfTweetsDeleted,noOfTweets+offset,idprint)
def deleteTweets(toDelete):
    progressInit(toDelete)
    for tid in toDelete:
        try:
            deleteTweet(tid)
        except Exception as errorOut:
            deleteFailed(errorOut,tid)
def getFileNames(dataFolder):
    tweetFileNames = glob.glob(dataFolder+"data/js/tweets/*")
    return tweetFileNames
def getTweetIds(tweetFileNames):
    tweetIDs = []
    for tweetFileName in tweetFileNames:
        tweetFile = open(tweetFileName)
        jsonInFile = tweetFile.read()
        tweetFileList = json.loads(jsonInFile[33:])
        for tweetJson in tweetFileList:
            tweetIDs.append(tweetJson["id"])
    return tweetIDs
main()